Your N and N+1 seem reasonable, but your data labeled N-1 graphs more like N-2. So the development time in between would be closer to N-1.
I suspect your camera test is giving you curves that include flare. This makes it difficult to fit to curves and come up with N development times.
If you have a curve family made of step wedge exposures that were contacted to film, you would have better developing time decisions.
But for now, you can use the times you found here, they certainly are good enough.